David Greenbaum, Director of UC Berkeley's Research IT, will provide an overview of national organizations that are supporting Research IT groups in higher ed, and what elements and initiatives may prove helpful to Berkeley's support of campus research. The overview will include a survey of current and developing organizations, and the changes they are responding to on campuses and nationally. “Established” research computing organizations include CASC, Educause/ECAR Cyberinfrastructure, XSEDE; developing efforts include ACI-REF, CaRC, and PEARC. David will focus on CaRC in particular as it is in a start up mode and is aiming to be quite broad and inclusive. Patrick Schmitz, Associate Director of Research IT and Program Manager of Berkeley Research Computing, will give an update on CASC, whose national meeting he attended in late March.
When: Thursday, April 6, 2017 from 12 - 1pm
Prior to the meeting, please review:
Presenting: David Greenbaum, Patrick Schmitz
Aaron Culich, Research IT
Chris Hoffman, Research IT
Deb McCaffrey, Research IT
Gary Jung, LBNL-HPCS & BRC
Jason Christopher, Research IT
John Lowe, Research IT
Kelly Rowland, Nuclear Engineering & BRC
Krishna Muriki, LBNL-HPCS & BRC
Maurice Manning, Research IT
Patrick Schmitz, Research IT
Quinn Dombrowski, Research IT
Rick Jaffe, Research IT
Steve Masover, Research IT
See slides (coming soon)
CI Framework: broader than Research IT or Research Computing.
[A survey of the alphabet soup of orgs: CaRC, CASC, Educause/ECAR, ACI-REF, XSEDE, Supercomputing, CNI, UC RITC, PEARC]
More not on the list on the slide: NERSC, Open Science Grid, PRP. Some deep in community building, others focused on a conference or on providing computational infrastructure. Focus on the slide's list is somewhat fuzzy, but these orgs are oriented to building community across EDU institutions & helping campus-level organizations like UCB's Research IT. A (non-canonical) list of activities in this vein is given in the slides.
CaRC is developing a survey (will be of ~150 institutions) to assess most useful, most common vision and purpose for the emerging organization (consortium) -- built from the contributed ideas of participants from 28 institutions currently involved in the development of CaRC. To be sustained by contributions from participating institutions.
ACI-REF may end up using CaRC as its long-term organizational home -- where it lives when it is no longer supported by the NSF grants on which the program was initially built.
CASC [Patrick; no slides]: Careful not to be a lobbying organization, but responds to requests from (for example) funding organizations (e.g., NSF, DoE, NIH). Meetings ~150 people, twice per year. Most meetings in DC because that's where the federal funders are; next meeting in Denver. Tends to draw folks who think about strategy and sustainability, as opposed to nuts-and-bolts people who run computational resources. Orgs participating include large institutions and small schools. Working groups (e.g., Protected Data; Beyond HPC). Group hovers between focus on best practices in research computing vs. focus on gov't funders. Member supported, $4,000 per year, which covers cost of conferences.
CASC v CaRC: is there a clear distinction, or is there significant overlap? What happens when some (many?) schools have to choose between a limited set of meetings to attend; and a limited amount of dues that can be paid out to this kind of organization. Activities take place on campuses, in contact between campuses. ACI-REF, CASC, XSEDE ... maybe PEARC as it develops? There's also annual meetings: PEARC, CASC ... different people have different perspectives about which of these are emerging, which are in full-flower, which are becoming moribund. Part of the reason for having this discussion is to raise awareness of the fact that these organizations &c. are in flux, and it's not clear whether and how each will emerge.
Chris notes that the draft paragraphs describing CaRC (on slide) mentioned HPC several times, a heavier focus than he has sensed in CaRC discussions he has been a part of. David notes that there's a tension between HPC (where many participants have been rooted) vs. other research IT needs emerging today ... the planned survey of ~150 universities will provide some interesting data about what people want the org to focus on.
Maurice and Quinn bring up the question of tools, codes, images, containers, documentation -- artifacts that fill the gap between computational infrastructure and the broad population of researchers who are not easily able to muck around in a Unix command line environment or write shell scripts; but with some higher level tools or scripts would be better able to utilize available resources.
Patrick: how does RDA (Research Data Alliance) fit in?
Chris: a whole side of this to do with research data and digital libraries -- some technically 'deep' participant individuals, but also government organizations, etc., who are functioning at more of a policy and strategy level. Research Data Access and Preservation is another network.
Rick: Maturity Model -- a deeper, more realistic way of thinking about how campuses evolve over time (a ten year plan, for example) to become able to deliver a full and mature program in support of research computation and/or research data management.
Chris: California version of ACI-REF -- a meeting in San Diego at the end of the month -- Maurice, Quinn, and Chris will participate
Maurice: need researchers to be incented to use expertise and resources that are provided by funders for more 'centralized' support as opposed to having a molecular biology PhD candidate spend her time coding Python that isn't really molecular biology research.
Patrick: an org focusing on Research Software Engineering in the UK. www.software.ac.uk